There are several aspects involved in delivering high I/O performance to parallel applications, from hardware characteristics to methods that manipulate workloads to improve achievable performance.

Running the same application with different I/O configurations gives the possibility to tune the I/O system according to the application access pattern.

One way to predict application performance in HPC systems with different I/O configurations is by using modelling and simulation techniques. Modeling the system allows assessing obtained performance and therewith estimate the performance potentially gained by optimizations.

File systems are implemented in the operating system which deploys strategies to improve performance such as scheduling, caching and aggregation. Therefore, the observable I/O performance depends on more than the capabilities of the raw block device.

• To develop general considerations about what influences the I/O performance.
• To analyze access pattern and define how it defines the performance of the I/O subsystems.
• To apply I/O strategies to improve the access pattern.
• To identify options for the deployed optimization strategies in a specific parallel file system.

• Select performance models to assess and optimize the application I/O performance.
• Identify tools capable of predicting the behavior of applications in HPC.
• Apply methods to manipulate workloads to improve achievable performance.